Preliminary Comparison of Polynomial Chaos and Splines for Uncertainty Quantification of Multihull Structural Loads

Abstract

The design and optimization of a ship’s structure depends on a prediction of lifetime extreme global loads. For novel hull forms, designers often rely on first-principle analysis and simulation methods for these predictions. Complicating this process is the fact that critical information about the ship, such as displacement and mass distribution, may be highly uncertain during early stage ship design. For this reason, efficient methods for uncertainty quantification of global structural loads are needed. This paper gives a preliminary comparison of two methods for this purpose polynomial chaos expansion, and multivariate splines. Preliminary results show favorable performance using multivariate splines. Interesting features of the design space are discussed, based on simulations using a scaled version of the Delft Catamaran 372 hull geometry.